It ain’t “round” if it’s missing \pi. BMI is “squareness”, with the \rm{height}^2 term, and BRI has to distinguish itself from that.
I think they’re modeling the body as an ellipsoid, so the 2\pi converts the waistline into a radius, and the half-height is the, uh, semi-major axis. Then they added a bunch of other fudge factors.
But that can’t possibly add to its predictive value since it’s just a fixed scaling factor.
Explain that to my wife when I leave \pi out of calculating pizza value!
As a researcher, I view changing the measure as a difficult hurdle. The new measure has to be much better. The old measure (BMI) already exists in every relevant dataset going back centuries. This is a huge advantage. Lots of literature exists related to BMI, which can give researchers a good sense of how it behaves with respect to other things you’re more interested in.
The main advantage of BRI, is that because it is relatively new, you can publish a bunch of papers redoing old BMI analyses with BRI instead. Unless BRI is capturing an important[1] amount of variance over BMI, it probably won’t gain much traction.
There’s also the triggering thing. People react poorly to BMI, so maybe BRI will make doctor’s jobs easier. Over half the people in the US would greatly benefit from exercising more and dropping 15-30 pounds, regardless of how it is quantified.
and important is different than significant ↩︎
The thing is, it’s not actually new. As Hari_Seldon noted, this is just a dressed-up waist-to-height ratio. It can’t be more predictive than that. And waist-to-height has already been analyzed a fair amount.